PANCHANGE_analysis
Katherine N Thompson, Christopher Hüebel, Molly R Davies, Kirstin L Purves
09 March, 2021
This is the analysis script to prepare the data to run analyses for the project “Anxiety, depression and trauma symptom change during the COVID-19 pandemic: retrospective versus objective assessment” - Young et al (2020)
Script written by K L Purves, K N Thompson, C Huebel and M Davies. Email: kirstin.purves@kcl.ac.uk, katie.thompson@kcl.ac.uk, christopher.1.huebel@kcl.ac.uk, molly.davies@kcl.ac.uk
#Set up
Clear global environment
Retrieve the current date to use it for file endings to not overwrite files when one exports files
#Packages Install packages (if they are not available in your version of R)
Load packages
Registered S3 method overwritten by 'pryr':
method from
print.bytes RcppFor best results, restart R session and update pander using devtools:: or remotes::install_github('rapporter/pander')
Attaching package: 'polycor'The following object is masked from 'package:psych':
polyserialcorrplot 0.84 loaded── Attaching packages ─────────────────────────────────────── tidyverse 1.3.0 ──✓ ggplot2 3.3.2 ✓ purrr 0.3.4
✓ tibble 3.0.4 ✓ dplyr 1.0.2
✓ tidyr 1.1.2 ✓ stringr 1.4.0
✓ readr 1.3.1 ✓ forcats 0.5.0── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
x ggplot2::%+%() masks psych::%+%()
x ggplot2::alpha() masks psych::alpha()
x dplyr::filter() masks stats::filter()
x dplyr::lag() masks stats::lag()
x tibble::view() masks summarytools::view()R functions
FunctionLib/lm_forest.R FunctionLib/lm_formula.R
value ? ?
visible FALSE FALSE
FunctionLib/lm_output.R FunctionLib/raincloud.R
value ? ?
visible FALSE FALSE Half-violin plot
Colour palettes
Define colours for plotting this are the standard coping colours
Choose in this chunk which palette to use 04.07.2020 - Default to 2 colour COPING palette
ggplot theme
Set up ggplot theme for the plots
Data import
Source data file paths
Read in data
[1] 34465 643Important vectors with collections of columns
Vectors for sum_score
[1] "gad.sum_score_prepan" "pcl.sum_score_prepan"
[3] "phq.sum_score_prepan" "phq.sum_score_8items_prepan"
[5] "ocir.sum_score_prepan" "gad.sum_score_base"
[7] "gad.sum_score_retro" "pcl.sum_score_base"
[9] "pcl.sum_score_retro" "phq.sum_score_base"
[11] "phq.sum_score_8items_base" "phq.sum_score_retro"
[13] "phq.sum_score_8items_retro" "ocir.sum_score_base"
[15] "ocir.sum_score_retro" [1] "phq.sum_score_prepan" "phq.sum_score_base" "phq.sum_score_retro" [1] "gad.sum_score_prepan" "gad.sum_score_base" "gad.sum_score_retro" [1] "ocir.sum_score_prepan" "ocir.sum_score_base" "ocir.sum_score_retro" [1] "pcl.sum_score_prepan" "pcl.sum_score_base" "pcl.sum_score_retro" Vectors for diff_score
[1] "phq.diff_score_retro_prepan" "phq.diff_score_base_prepan"
[3] "phq.diff_score_base_retro" "gad.diff_score_retro_prepan"
[5] "gad.diff_score_base_prepan" "gad.diff_score_base_retro"
[7] "pcl.diff_score_base_prepan" "ocir.diff_score_retro_prepan"
[9] "ocir.diff_score_base_prepan" "ocir.diff_score_base_retro" [1] "phq.diff_score_retro_prepan" "phq.diff_score_base_prepan"
[3] "phq.diff_score_base_retro" [1] "gad.diff_score_retro_prepan" "gad.diff_score_base_prepan"
[3] "gad.diff_score_base_retro" [1] "ocir.diff_score_retro_prepan" "ocir.diff_score_base_prepan"
[3] "ocir.diff_score_base_retro" Create subsets of data for analyses
Filter data frame to create 2 additional datasets. Additional exclusion criteria (beyond those listed above) are specified below:
- Retrospective - baseline change anlyses
- Exclude individuals missing retrospective and/or baseline data
- Prepandemic - baseline change analyses i. Exclude individuals missing pandemic and/or baseline data ii. Exclude individuals with prepandemic data collected after 1st March.
Retro-baseline analyses
Note: make sure we impute retrospective data for those who did not change their feelings
Important note::: PCL does not have retrospective data, so we allow Baseline only or All prepandemic, baseline and retrospective OR Baseline and prepandemic data. Important logic here is that we need EVRYONE with ANY baseline data, but dont care about any other time point for this retrospective dataset for PCL.
[1] 32737 643Prepan-baseline analyses Exclude individuals with NA for MHD, age, gender; exclude if prepan data collected after 1 March
[1] 12157 643Prepan-baseline analyses Exclude individuals with NA for MHD, age, gender; exclude if prepan data collected after 1 March (to ensure comparability with prepan analyses)
[1] 10489 643age^2 sensitivity Exclude individuals with NA for continuous Age variable
[1] 25814 643[1] 24096 643[1] 12157 643Sensitivity analysis for individuals who signed up to GLAD (prepan) either 3, 6 or 9 months before the pendemic began.
#ANALYSIS PLAN 1. Calculate mean change per disorder group i. Disorders (9) a. Rank the disorders by the size of the change ii. Sex iii. Age (13; RAMP; potentially collapse above 65 depending on group sizes) iv. Ethnicity groups (5; White European, Black (British), Asian, Arab, Mixed)
1. Questionnaire comparions: Calculate mean change per group
gad
Summary scores
Warning in min(., na.rm = T): no non-missing arguments to min; returning InfWarning in max(., na.rm = T): no non-missing arguments to max; returning -InfTable for knitting gad sum score summary table
| Change | Variable | Min | Max | Mean | Median | SD | Skew | Kurtosis |
|---|---|---|---|---|---|---|---|---|
| No change | Base | 0 | 21 | 2.740042 | 1 | 4.481334 | 2.1348352 | 4.2519780 |
| No change | Prepan | 0 | 21 | 7.993625 | 7 | 6.165962 | 0.5274724 | -0.8425302 |
| No change | Retro | 0 | 21 | 2.740042 | 1 | 4.481334 | 2.1348352 | 4.2519780 |
| Worsening | Base | 0 | 21 | 8.766138 | 7 | 5.693132 | 0.5145419 | -0.8093878 |
| Worsening | Prepan | 0 | 21 | 9.160358 | 8 | 5.869416 | 0.3772128 | -0.9054598 |
| Worsening | Retro | 0 | 21 | 4.658218 | 4 | 4.263868 | 1.3773597 | 1.9563781 |
| Improvement | Base | 0 | 21 | 4.767912 | 3 | 4.832619 | 1.3721605 | 1.5870653 |
| Improvement | Prepan | 0 | 21 | 9.046937 | 8 | 5.804418 | 0.3595536 | -0.8773999 |
| Improvement | Retro | 0 | 21 | 8.337588 | 7 | 5.489836 | 0.5192540 | -0.5883912 |
| NA | Base | 0 | 21 | 5.722222 | 5 | 5.816102 | 1.0412406 | 0.2319193 |
| NA | Prepan | 0 | 21 | 10.000000 | 13 | 8.563488 | -0.0313927 | -1.9895986 |
| NA | Retro | Inf | -Inf | NaN | NA | NA | NA | NA |
Differences scores
Warning in min(., na.rm = T): no non-missing arguments to min; returning Inf
Warning in min(., na.rm = T): no non-missing arguments to min; returning Inf
Warning in min(., na.rm = T): no non-missing arguments to min; returning Inf
Warning in min(., na.rm = T): no non-missing arguments to min; returning InfWarning in max(., na.rm = T): no non-missing arguments to max; returning -Inf
Warning in max(., na.rm = T): no non-missing arguments to max; returning -Inf
Warning in max(., na.rm = T): no non-missing arguments to max; returning -Inf
Warning in max(., na.rm = T): no non-missing arguments to max; returning -InfTable for knitting gad diff summary table
| Change | Variable | Min | Max | Mean | Median | SD | Skew | Kurtosis |
|---|---|---|---|---|---|---|---|---|
| No change | Base - prepan | -21 | 20 | -2.0355457 | -1 | 4.695963 | -0.4873054 | 1.5249971 |
| No change | Base - retro | Inf | -Inf | NaN | NA | NA | NA | NA |
| No change | Prepan - retro | Inf | -Inf | NaN | NA | NA | NA | NA |
| Worsening | Base - prepan | -18 | 21 | 1.0244339 | 1 | 5.272565 | -0.0232123 | 0.6264062 |
| Worsening | Base - retro | -21 | 21 | 4.1083717 | 3 | 4.319494 | 0.7492612 | 0.9033930 |
| Worsening | Prepan - retro | -21 | 18 | -3.4649078 | -3 | 5.071721 | -0.3175882 | 0.4750981 |
| Improvement | Base - prepan | -21 | 12 | -3.9601911 | -3 | 5.470385 | -0.2955275 | 0.1734936 |
| Improvement | Base - retro | -21 | 21 | -3.5651408 | -3 | 5.333376 | 0.4506964 | 2.0385927 |
| Improvement | Prepan - retro | -21 | 20 | 0.4334661 | 1 | 5.679098 | -0.1488051 | 0.8650700 |
| NA | Base - prepan | -9 | 6 | -1.5000000 | -2 | 5.431390 | 0.0468089 | -1.6798094 |
| NA | Base - retro | Inf | -Inf | NaN | NA | NA | NA | NA |
| NA | Prepan - retro | Inf | -Inf | NaN | NA | NA | NA | NA |
Long format
Plots diff
makes this into percentages
Warning: Removed 60925 rows containing non-finite values (stat_count).
Warning: Removed 35523 rows containing non-finite values (stat_bin).
Warning: Removed 35523 rows containing non-finite values (stat_bin).
Warning: Removed 25402 rows containing non-finite values (stat_bin).
Plots sum
Warning: Removed 21890 rows containing non-finite values (stat_bin).
Warning: Removed 21890 rows containing non-finite values (stat_bin).
Violin plots
Warning: `fun.y` is deprecated. Use `fun` instead.Warning: `fun.ymin` is deprecated. Use `fun.min` instead.Warning: `fun.ymax` is deprecated. Use `fun.max` instead.Warning: Removed 21890 rows containing non-finite values (stat_ydensity).Warning: Removed 21890 rows containing non-finite values (stat_summary).Warning: Removed 21890 rows containing missing values (geom_point).
Warning: `fun.y` is deprecated. Use `fun` instead.Warning: `fun.ymin` is deprecated. Use `fun.min` instead.Warning: `fun.ymax` is deprecated. Use `fun.max` instead.Warning: Removed 21890 rows containing non-finite values (stat_ydensity).Warning: Removed 21890 rows containing non-finite values (stat_summary).Warning: Removed 21890 rows containing missing values (geom_point).
Warning: `fun.y` is deprecated. Use `fun` instead.Warning: `fun.ymin` is deprecated. Use `fun.min` instead.Warning: `fun.ymax` is deprecated. Use `fun.max` instead.Warning: Removed 21890 rows containing non-finite values (stat_ydensity).Warning: Removed 21890 rows containing non-finite values (stat_summary).Warning: Removed 21890 rows containing missing values (geom_point).
Warning: `fun.y` is deprecated. Use `fun` instead.Warning: `fun.ymin` is deprecated. Use `fun.min` instead.Warning: `fun.ymax` is deprecated. Use `fun.max` instead.Warning: Removed 21890 rows containing non-finite values (stat_ydensity).Warning: Removed 21890 rows containing non-finite values (stat_summary).Warning: Removed 21890 rows containing missing values (geom_point).